Page 2 of 2
Re: Spectral Analysis math
Posted: Tue May 21, 2013 7:46 am
by jean-eric
Do you know the convention for the Fourier Transform parameters (a and b just below) used ?
1/n^((1-a)/2)Underoverscript[[Sum], r=1, n]Subscript[u, r] e^(2[Pi] i b(r-1)(s-1)/n)
Re: Spectral Analysis math
Posted: Thu May 23, 2013 10:42 am
by jean-eric
No news to answer to the question pf the FFT convbentions made by Audacity ?
Well I do not manage to match Mathematica 9 to the Audacity results but I guess this is a convention missmatch more than a bug.
Re: Spectral Analysis math
Posted: Thu May 23, 2013 11:46 am
by Robert J. H.
Could you explain where this "convention" originates and what is actually meant by it?
We are no mathematicians.
Re: Spectral Analysis math
Posted: Thu May 23, 2013 7:40 pm
by jean-eric
Well, at least I know this kind of things
the Fourier Transform is something involving coefficient of
Exp with i^2=-1
but
the argument of the exponentiel may be defined differently and this influence the normalization of the function too.
There is of course a set of definition that garantee the fact that InverseFourier[Fourier] = Identity
but from a set of parameter to another two Fourier Transform might give different numerical results.